In large-scale multiplayer titles where there are a considerable number of actors and components that need to be updated, game logic can quickly become the bottleneck. This is especially true server side and on performance-limited platforms.
In this session, Daniel Rätzer presents an approach to writing game code that increases data and instruction cache coherency, as well as allowing for automatic parallelization to maximize thread occupancy. Daniel talks about ECS in general, and how Funcom implemented a scheduler which extends on the taskgraph and that can dispatch and batch process game logic on sets of components.
Great for: engine and game programmers working on titles where the game thread is the bottleneck.
Learn more at http://www.unrealengine.com/games
source
